How 99% Reconciliation Accuracy Raises the Bar for MCA Document Verification

Reconciliation Accuracy Just Hit a New High. Now What?

LendPathway recently announced that 99.7% of all ledgers processed through its platform now reconcile correctly, meaning starting balance plus transactions equals ending balance with mathematical precision. The announcement, reported by deBanked in March 2026, signals that automated bank statement analysis for lenders has crossed a critical threshold. For MCA funders and alternative lenders, this is both encouraging and incomplete.

Encouraging because reconciliation is one of the most tedious, error-prone steps in underwriting. When a platform can verify that the math on a bank statement or ledger adds up with near-perfect reliability, underwriters can redirect hours of manual checking toward higher-value judgment calls. Incomplete because reconciliation accuracy answers only one question: do the numbers on this document add up? It does not answer the more dangerous question: is this document real?

That distinction matters enormously. Fraudulent bank statements are often internally consistent. A well-crafted fake will reconcile perfectly because the forger controls every number on the page. The real challenge for MCA lenders is not whether a document's math checks out. It is whether the document itself is authentic. This article breaks down what the new accuracy benchmarks mean for your underwriting workflow, where the gap still exists, and how to close it.

What 99%+ Reconciliation Accuracy Actually Means for MCA Lenders

Speed Gains in Document Processing

At 99.7% accuracy, a reconciliation engine can process hundreds of bank statements per day without generating a backlog of exceptions that require human review. For MCA operations teams, this translates directly into faster deal velocity. Instead of an underwriter spending 20 to 30 minutes per file checking whether deposits, withdrawals, and balances align, the system flags only the fraction of a percent that fails reconciliation. The math layer of underwriting becomes nearly autonomous.

This is a meaningful operational improvement. In a high-volume shop processing 50 or more applications per day, eliminating manual reconciliation checks can free up an entire underwriter's worth of capacity. That capacity can be redirected toward analyzing cash flow patterns, evaluating revenue consistency, or reviewing flagged accounts in depth.

The Limits of Mathematical Verification

Reconciliation accuracy, however sophisticated, operates on the data it receives. It confirms internal consistency. It does not confirm external validity. Consider three common fraud scenarios that pass reconciliation with flying colors.

First, a borrower uses PDF editing software to inflate deposit amounts across three months of statements. Every page balances. Every running total is correct. The reconciliation engine processes the file and returns a clean result because the forger did the arithmetic properly.

Second, an applicant submits statements from a secondary account that has been temporarily inflated through circular transfers. The deposits are real. The balances are real. Reconciliation succeeds. But the underlying cash flow is manufactured.

Third, a broker submits documents on behalf of a merchant using a template that mirrors the formatting of a major Canadian bank. The numbers are fabricated but consistent. The system processes them as valid because nothing fails the math check.

In each case, the weakness is not in the reconciliation engine. It is in the assumption that a mathematically correct document is a trustworthy one. As we explored in our analysis of how MCA lenders detect synthetic identity fraud in bank verification, document-level fraud often passes automated checks precisely because the forger builds the document to satisfy them.

The Authenticity Verification Gap

This is the gap that the industry's push toward automation has not fully closed. Reconciliation platforms, OCR engines, and AI-powered document classifiers are getting remarkably good at reading and interpreting financial documents. What they cannot do is confirm that the document they are reading was generated by a real bank, from a real account, showing real transaction history.

Closing that gap requires a different kind of verification altogether. It requires seeing the data at its source. Not a PDF that could have been edited. Not a screenshot that could have been composited. A live view of the banking portal itself, captured in real time, showing the same numbers the underwriter will later analyze.

Pairing Automation with Visual Bank Verification

The most effective underwriting workflows in 2026 are not choosing between automation and manual verification. They are layering both. Automated reconciliation handles the math. Visual verification handles the trust.

How Async Recording Fills the Gap

Exact Balance sits at the intersection of these two needs. When a lender sends a verification request through the platform, the applicant receives a secure link and records their live banking session directly in their browser. No software installation. No scheduled call. The applicant logs into their actual bank portal, navigates through the account summary, transaction history, and any specific date ranges the underwriter has requested, and the entire session is captured as a timestamped video.

The underwriter then reviews the recording on their own schedule. They can see the bank's URL in the browser bar. They can watch the applicant scroll through real transaction data. They can confirm that the balances shown in the live portal match the balances on the submitted statements. If the numbers diverge, that is an immediate red flag that no reconciliation engine would catch on its own.

This approach solves several problems simultaneously. It verifies that the document source is legitimate. It creates a visual audit trail that holds up under scrutiny. And it does all of this asynchronously, meaning neither the underwriter nor the applicant needs to coordinate a live call. As we covered in our breakdown of why screen recording beats live verification calls for MCA lenders, removing the scheduling bottleneck is one of the highest-leverage improvements a lending operation can make.

AI-Guided Session Validation

Exact Balance's AI-guided recording feature adds another layer of reliability. A floating coach walks the applicant through each required step, confirming in real time that they have shown the correct screens, navigated to the right date ranges, and displayed the necessary account details. This is not a passive screen capture tool. The AI validates completion as the session progresses, reducing the number of incomplete or unusable recordings that waste underwriter time.

On the review side, activity tracking logs show exactly when the link was opened, when recording started, and when the submission was completed. Combined with the video itself, this creates a full audit trail that pairs naturally with the reconciliation data from automated document processing platforms.

Building a Layered Verification Workflow

A practical workflow looks like this. The applicant submits bank statements, either directly or through a broker. Those statements run through an automated reconciliation and data extraction engine. Simultaneously, or immediately after, the lender sends an Exact Balance verification request. The applicant records their banking portal. The underwriter reviews both the extracted data and the video recording side by side.

If the reconciled numbers from the document match what appears in the live banking session, confidence is high. If there is any discrepancy, the underwriter has clear, visual evidence to support a decline or a request for additional documentation. This layered approach catches fraud that pure automation misses while maintaining the speed advantages that automation provides.

What This Means for MCA Operations in Practice

Upstart CEO Paul Gu made a striking comment during the company's Q4 earnings call, noting that humans have never been very good at precisely underwriting loans, and that AI inherits many of the same limitations. The observation is more nuanced than it sounds. AI excels at pattern recognition, speed, and consistency. Humans excel at judgment, context, and detecting when something feels wrong even if the data looks clean.

The best MCA underwriting operations in 2026 leverage both. They use automation to handle volume and flag anomalies. They use human reviewers, supported by tools like async video verification, to make final trust decisions on flagged files. They do not treat automation as a replacement for verification. They treat it as the first pass in a multi-layered process.

For Canadian MCA lenders specifically, the stakes are rising. As we detailed in our coverage of common mistakes MCA companies make with bank verification early on, many growing operations skip visual verification entirely because they believe automated document checks are sufficient. That assumption holds until the first well-crafted fraud slips through.

Consider the economics. A single funded deal based on fabricated bank statements can cost a lender anywhere from $25,000 to $250,000 or more. The cost of adding async visual verification to every deal is a fraction of that. At Exact Balance's Basic tier, 250 verifications per month costs $1,750, which works out to $7 per verification. That is a negligible line item compared to the loss exposure of even one fraudulent advance.

The operational math is equally compelling. Without async verification, an underwriter might spend 30 to 45 minutes coordinating and conducting a live verification call. With Exact Balance, the applicant records on their own time, and the underwriter reviews the recording in under 10 minutes. Multiply that time savings across dozens of daily applications and the capacity gains are significant.

Frequently Asked Questions

What is reconciliation accuracy in automated bank statement analysis?

Reconciliation accuracy measures whether an automated system can correctly verify that a bank statement's starting balance, plus deposits, minus withdrawals, equals the ending balance. A 99.7% accuracy rate means that nearly every document processed passes this mathematical consistency check without manual intervention. While high reconciliation accuracy dramatically reduces processing time, it does not verify whether the document itself is authentic or unaltered.

Can fraudulent bank statements pass automated reconciliation?

Yes. A well-fabricated bank statement will often pass reconciliation checks because the forger ensures that all numbers add up correctly. Automated reconciliation confirms internal mathematical consistency, not external authenticity. This is why leading MCA lenders supplement automated document analysis with visual verification of live banking sessions, confirming that the data in submitted documents matches what the bank portal actually shows.

How does async bank verification work for MCA lenders?

Async bank verification eliminates the need for live, scheduled phone calls between underwriters and applicants. Instead, the lender sends the applicant a secure link. The applicant opens the link, records their live banking session directly in their browser, and submits the recording. The underwriter reviews the video on their own schedule, verifying account details, transaction history, and balances without any real-time coordination. Exact Balance provides this workflow with AI-guided recording, activity tracking, and encrypted cloud storage.

Should MCA lenders use both automated analysis and visual verification?

Absolutely. Automated bank statement analysis handles speed and mathematical consistency at scale. Visual verification through recorded banking sessions handles authenticity and trust. Used together, they create a verification workflow that is both fast and fraud-resistant. The automated layer processes volume efficiently, while the visual layer catches the sophisticated fraud that document-level checks miss.

Conclusion

The alternative lending industry's push toward automated document processing is producing real results. Reconciliation accuracy above 99% is no longer aspirational; it is the new baseline. But accuracy on a document and trust in a document are two different things, and MCA lenders who conflate the two are exposed to exactly the kind of fraud that automation cannot catch on its own.

The answer is not to choose between automation and verification. It is to use both. Let automated platforms handle the math. Let visual, async verification confirm the source. Together, they create an underwriting workflow that is fast, scalable, and resilient against even sophisticated document fraud.

Visit exactbalance.ca to see how async bank verification fits into your underwriting workflow and closes the gap between document accuracy and document authenticity.